SEAL: Towards Safe Autonomous Driving via Skill-Enabled Adversary Learning for Closed-Loop Scenario Generation

TL;DR

SEAL introduces a novel scenario generation method for autonomous driving that uses learned objectives and human-like adversarial skills to create more realistic and diverse safety-critical scenarios, improving system robustness.

Contribution

The paper presents SEAL, a new approach for generating realistic adversarial scenarios using learned objectives and skills, enhancing safety validation for autonomous driving systems.

Findings

Generated scenarios are more realistic than state-of-the-art baselines.

Improved ego task success rate by over 20% across various scenarios.

Facilitates future research with released code and tools.

Abstract

Verification and validation of autonomous driving (AD) systems and components is of increasing importance, as such technology increases in real-world prevalence. Safety-critical scenario generation is a key approach to robustify AD policies through closed-loop training. However, existing approaches for scenario generation rely on simplistic objectives, resulting in overly-aggressive or non-reactive adversarial behaviors. To generate diverse adversarial yet realistic scenarios, we propose SEAL, a scenario perturbation approach which leverages learned objective functions and adversarial, human-like skills. SEAL-perturbed scenarios are more realistic than SOTA baselines, leading to improved ego task success across real-world, in-distribution, and out-of-distribution scenarios, of more than 20%. To facilitate future research, we release our code and tools: https://github.com/cmubig/SEAL